A Mine Plow System Neural Network Modeling And Control

Neural networks have been proven to be a successfully model of complex system with outstanding abilities in nonlinear mapping, self learning and etc.Firstly, the electrohydraulic servo system of a certain mine sweeping plough is introduced, and how to obtain the input-output data is proposed.Then, several methods, including the first-principle method and intelligent method, are detailed to construct different model of the electrohydraulic servo system. Based on analysis of the identified five models, the neural network is chosen as the most powerful tool for modelling of the electrohydraulic servo system.On the basis of above-mentioned works, this paper researched on modelling of the electrohydraulic servo system based on neural networks. The two most common types of neural networks, i.e. multilayer perceptron (MLP) neural networks and radial basis function (RBF) neural networks, were discussed, and their four typical training algorithms were investigated. The results indicated that the RBF neural network is appropriate for modelling of this system.Two methods based on genetic algorithm are proposed for designing the RBF neural network. One method for identifying parameters of the RBF neural network is proposed in this dissertation. There are two stages for optimizing the RBF neural network: structure optimization stage and parameters optimization stage, so this paper proposed another method for designing the RBF neural network based on hierarchical genetic algorithm (HGA) is also proposed. The results clearly showed validity of these two methods.Design of RBF neural network is a typical multi-objective optimization problem, so a method of designing RBF neural network based on Non-dominated Sorting Genetic Algorithm-II (NSGA-II) and HGA is proposed. The proposed method is applied to the modelling of the electrohydraulic system, and the results clearly indicate its validity.Finally, this dissertation designed the neural network based direct inverse controller (NNDIC) of the electrohydraulic servo system with the obtained neural network. The experimental results validated the feasibility of the modelling and control methods.